cluster_analytics

Input data

Three data set are used: HRSL OSM *Admin boundaries

#read in spatial data

datapath<-("\\\\srv02\\RL-Institut\\04_Projekte\\240_NESP2\\03-Projektinhalte\\02_Data\\02_Data_from_online_sources\\05_Population_data\\hrsl_nga_v1")
builtup<-raster(file.path(datapath,"hrsl_nga_settlement.tif"))
adm1<-st_read("\\\\srv02\\RL-Institut\\04_Projekte\\240_NESP2\\03-Projektinhalte\\02_Data\\02_Data_from_online_sources\\04_Administrative_Boundaries\\nga_admbnda_adm1_osgof")

## Reading layer `nga_admbnda_adm1_osgof_20161215' from data source `\\srv02\RL-Institut\04_Projekte\240_NESP2\03-Projektinhalte\02_Data\02_Data_from_online_sources\04_Administrative_Boundaries\nga_admbnda_adm1_osgof' using driver `ESRI Shapefile'
## Simple feature collection with 37 features and 12 fields
## geometry type:  MULTIPOLYGON
## dimension:      XY
## bbox:           xmin: 2.668534 ymin: 4.273007 xmax: 14.67882 ymax: 13.89442
## epsg (SRID):    4326
## proj4string:    +proj=longlat +datum=WGS84 +no_defs

adm1 %>% select(admin1Name) %>% ggplot() + geom_sf() + labs(title = "Nigeria")

adm1 %>% filter(admin1Name == "Kano") %>% ggplot() + geom_sf() + labs(title = "Kano")

state_list<-list()

Data processing

Extract the population raster data for each state

Clustering

The DBSCAN algorithm is used to cluster the points based on the distance to each other and a minimum number of points per cluster.

test=dbscan(input, eps = 0.00083, minPts = 3, weights = NULL)
input$cluster <- test$cluster

#split the original data into two according to whether dbscan has assigned or cluster or noise.

groups  <- input %>% filter(cluster != 0)
noise  <- input %>% filter(cluster== 0)

out <- input %>% 
  st_as_sf(coords = c("x","y"), crs = 4326) %>%
  group_by(cluster) %>% 
  summarise() %>% 
  st_convex_hull()

out %>% filter(cluster == 0) 

## Simple feature collection with 1 feature and 1 field
## geometry type:  POLYGON
## dimension:      XY
## bbox:           xmin: 7.688408 ymin: 10.54979 xmax: 9.363963 ymax: 12.59645
## epsg (SRID):    4326
## proj4string:    +proj=longlat +datum=WGS84 +no_defs
## # A tibble: 1 x 2
##   cluster                                                          geometry
##     <int>                                                    <POLYGON [°]>
## 1       0 ((8.74563 10.54979, 8.736185 10.55062, 8.716463 10.55367, 8.5495~

outfiltered <- out %>% filter(cluster != 0)
union <- st_union(outfiltered)

Visualization

The resulting clusters are shown in the map below.

powerlines<-st_read("\\\\srv02\\RL-Institut\\04_Projekte\\240_NESP2\\03-Projektinhalte\\02_Data\\02_Data_from_online_sources\\06_power_infrastructure\\columbia_grid_tracking\\kedco-grid-data-2016-shapefiles\\KEDCO_Grid_Data_Simplified_06172015_with_overlaps_removed.shp")

## Reading layer `KEDCO_Grid_Data_Simplified_06172015_with_overlaps_removed' from data source `\\srv02\RL-Institut\04_Projekte\240_NESP2\03-Projektinhalte\02_Data\02_Data_from_online_sources\06_power_infrastructure\columbia_grid_tracking\kedco-grid-data-2016-shapefiles\KEDCO_Grid_Data_Simplified_06172015_with_overlaps_removed.shp' using driver `ESRI Shapefile'
## Simple feature collection with 5022 features and 8 fields
## geometry type:  LINESTRING
## dimension:      XY
## bbox:           xmin: 6.988658 ymin: 9.23255 xmax: 12.45614 ymax: 13.32508
## epsg (SRID):    4326
## proj4string:    +proj=longlat +datum=WGS84 +no_defs

mapview(st_geometry(powerlines$geometry))+
mapview(union[[1]], color= "red", col.regions = "red")

# Transform the sfc object to polygons

#poly <- st_collection_extract(union, type = "POLYGON", warn = FALSE)
#mapview(poly, color= "red", col.regions = "red")+
 # mapview(st_geometry(powerlines$geometry))




#Calculate centroids of the polygons:

#poly$centroids <-
  #st_centroid( x = poly ) %>%
 # st_geometry()

Distance calculations

Between the polygons and grid infrastructure

#powerlines<-st_read("\\\\srv02\\RL-Institut\\04_Projekte\\240_NESP2\\03-Projektinhalte\\02_Data\\02_Data_from_online_sources\\06_power_infrastructure\\columbia_grid_tracking\\kedco-grid-data-2016-shapefiles\\KEDCO_Grid_Data_Simplified_06172015_with_overlaps_removed.shp")

# Calculate distances -----------------------------------------------------

#dist <- st_distance(centroids, st_cast(st_geometry(powerlines), "POINT"), by_element = TRUE)